Page 1 of 2

Finally understanding knock / detonation detection on ca18det

Posted: Mon May 11, 2020 10:05 am
by Shaker
Hello there.

After reading several posts about detonation detection on a CA18DET and because I have to somehow monitor knock on my current setup, I have a few questions:

1) Is there still something like a "knock map" in the nistune software of the Type 1 board? I've read about it, but have never seen it. Is that an old feature, now obsolete and removed?
1a) If not obsolete, where do I find the knock maps?
1b) If yes, how does nistune deal with knock, if it occurrs?

2) The factory strategy to retard timing by 1.2 degrees per detonation (info from technical manual), is this still active after having the nistune board installed?
2a) If not, see 1b)

3) After engine rebuild, I've had to deal with knock for some reason, which is audible from the inside (this irregularly clicking, rattling noise). But there is no knock reported in nistune, not one single time. Even worse: The ECU doesn't seem to do anything against this knock. I had to retard timing manually in the nistune timing map, which helped. But now I'm unsure of course, if and how the nistune ECU can protect the engine.
From what you know from analyzing these ECUs, is it possible that the detonation detection board is broken but reporting to be ok to the rest of the ECU or the DTC system?

Signatures seem to be deactivated. My ECU is: 23710-39F00, Type 1 V5 board, 39F0F base image. Nistune 1.3.7

Re: Finally understanding knock / detonation detection on ca18det

Posted: Mon May 11, 2020 2:43 pm
by Matt
1. There are low octane knock maps which the ECU used to switch too when excessive knock was recorded. It would stick to those maps until the vehicle as restarted. Those were removed and replaced with flex maps in feature pack 1 firmware
a. Removed. Only going back to standard Nissan firmware (say 39F00 instead of 39F0F). Tuned cars run high octane fuel so the need to use that map was obsolete. Most tuners make both maps the same (knock and main) so it made it redundant
b. It retards timing on knock. Keeps track of knock counts and pulls out timing. It does not switch over

2. Yes

3. The factory knock sensing stuff is unchanged. If there is no 'knock' window available and showing knock then it is not available. In feature pack there is no knock reporting (I'm working on an improved version at the moment)

The main thing to watch is your TRACE window for 'igniton timing' and see if timing is retarded compared to the timing map

If it is then the ECU is pulling knock, otherwise it is either not detecting or picking up on it. Check with knock headphones for when it is chattering

Re: Finally understanding knock / detonation detection on ca18det

Posted: Mon May 11, 2020 8:41 pm
by Shaker
Thanks for your explanations!
These flex maps only apply when using the flex fuel feature, right? So if I am not using flex fuel, for me the stock knock handling should be performed, correct?

But what if that doesn't happen? Can the board be broken but reports being ok? So no DTC would be set.

What exactly do you mean "it's not available"? I have the "knock warning panel", which used to show random values when uploading a map even if the engine was off (you remember maybe?). In version 1.3.7 this window doesn't show anything even if I hear knock. Are you saying, the feature pack version was never meant to display knock though?

And by "trace window" you mean the window that opens on the button labeled "TRACE"? Actually I've never used that, but I will have a look what it does.

I wonder, what a detonation detection board is good for, if it doesn't react on knock I can even hear with my ears while driving...

In which range (rpm restricted?) can this ECU detect knock? And does nistune when installed change or enhance this capability?

Re: Finally understanding knock / detonation detection on ca18det

Posted: Tue May 12, 2020 11:46 pm
by Matt
These flex maps only apply when using the flex fuel feature, right? So if I am not using flex fuel, for me the stock knock handling should be performed, correct?
Modifying the ECU code is complicated, so the original reference to the knock maps were removed. Too many tuners getting confused when it was/wasn't switching to the knock map made it more complicated than say a power FC, Haltec etc which just has one set of maps
But what if that doesn't happen? Can the board be broken but reports being ok? So no DTC would be set.
If there is a knock fault code (disconnected for example) then knock fault is raised, and timing is retarded normally 5 degrees extra
What exactly do you mean "it's not available"? I have the "knock warning panel", which used to show random values when uploading a map even if the engine was off (you remember maybe?).
When I was looking at the firmware for RB30 and RB20 I noticed that the knock counting firmware was not enabled when the FP builds were built (it is separate code). I would need to double check if CA18DET was affected, maybe it wasn't. I would need to check again
In version 1.3.7 this window doesn't show anything even if I hear knock. Are you saying, the feature pack version was never meant to display knock though?
Knock reporting was added before feature pack for about 5 different ECUs. Recently I've been doing updates (adding extra knock reporting per cylinder) and noticed it was not reporting updates for the RB20/RB30 and reporting had been disabled.

It was still meant to display knock however, so that will be fixed in an update.

S13 CA18DET I need to check again, but likely the next update will add per cylinder knock reporting similar to Z32 VG30/BNR32 VG30 ECUs which I've been working on (also for R34 RB25DET)

Keep in mind, that I'm only talking about the reporting of the main knock count in the ECUs, not the effects of knock (which should retard timing, and still functions as normal)
And by "trace window" you mean the window that opens on the button labeled "TRACE"? Actually I've never used that, but I will have a look what it does.
Definately use it. You can see when it retards timing on knock and compare against your timing map.
I wonder, what a detonation detection board is good for, if it doesn't react on knock I can even hear with my ears while driving...
The board monitors the knock sensor for frequencies where the knock occurs. It is a narrowband filter, and it sometime can pick up engine noise (incorrectly as knock)

If the load scales have been moved around in the tune, or the knock cells disabled that may affect it retarding timing. What is your ECU part number with Feature pack? I can check the firmware on the bench for knock count
In which range (rpm restricted?) can this ECU detect knock? And does nistune when installed change or enhance this capability?
None of this is changed with Nistune fitted (only low octane map removed). The range is where the flags are in the timing maps. During normal operation (assuming your knock sensor is working) it should retard timing when knock is detected.

Re: Finally understanding knock / detonation detection on ca18det

Posted: Wed May 13, 2020 5:35 am
by Shaker
Wow, that's a lot to understand, mostly because of all the different terms...
So, is the "firmware" of the board, what is sometimes referred to as "base image"?
Matt wrote: Tue May 12, 2020 11:46 pm If there is a knock fault code (disconnected for example) then knock fault is raised, and timing is retarded normally 5 degrees extra
I mean, what if the detonation detection board reports to the main ECU board, that everything is ok, but the board self cannot detect knock anymore for some reason? Then no DTC would be set and no actions would be taken.
Matt wrote: Tue May 12, 2020 11:46 pm When I was looking at the firmware for RB30 and RB20 I noticed that the knock counting firmware was not enabled when the FP builds were built (it is separate code). I would need to double check if CA18DET was affected, maybe it wasn't. I would need to check again

Knock reporting was added before feature pack for about 5 different ECUs. Recently I've been doing updates (adding extra knock reporting per cylinder) and noticed it was not reporting updates for the RB20/RB30 and reporting had been disabled.

It was still meant to display knock however, so that will be fixed in an update.
How could it display knock if the knock count is accidentally disabled in some boards?
Matt wrote: Tue May 12, 2020 11:46 pm If the load scales have been moved around in the tune, or the knock cells disabled that may affect it retarding timing. What is your ECU part number with Feature pack? I can check the firmware on the bench for knock count
Could you please explain how to enable or disable knock cells? What does that mean?
In the context menu of the timing map I have "Highlight Knock Region C" disabled and "Auto update knock maps", which confuses me again, since you say, knock maps have been removed. :?: (That's on my laptop I use to watch logs with more comfort. It has never been connected to the ECU...)

My ECU number and the base image version are mentioned in my first post: 23710-39F00, Type 1 V5 board, 39F0F base image. Is that what you need?
Matt wrote: Tue May 12, 2020 11:46 pm None of this is changed with Nistune fitted (only low octane map removed). The range is where the flags are in the timing maps. During normal operation (assuming your knock sensor is working) it should retard timing when knock is detected.
I've heared that the old ECU has some limitations like it cannot detect knock above 4000 rpm, just because of the lack of computing power. That's why I'm asking...

And because I am missing an answer and to be sure: The whole Flex Fuel area - can this be ignored by me if I do not use the Flex Fuel feature?

Another thing: Is it normal that the green selection stays where it is in the Knock Retard Limit table? Even though I'm moving around in the log? Shouldn't it be moving, too according to rpm and load?
Nistune 1.png
(748.14 KiB) Downloaded 2062 times
Nistune 2.png
(747.3 KiB) Downloaded 2062 times
Nistune 3.png
(747.77 KiB) Downloaded 2062 times

Re: Finally understanding knock / detonation detection on ca18det

Posted: Sun May 17, 2020 8:35 pm
by Matt
Long post, and a lot of questions. So I had to collect some info and run on the bench before replying
So, is the "firmware" of the board, what is sometimes referred to as "base image"
Nissan ROM file we download/read is the factory 'base image' and we patch it with the Nistune 'firmware'
The original firmware was just to add consult to the board (Type 1) via USB, and to read/write the ECU
Later updates to the original firmware added a knock count (incremented on knock), and knock report command (reported knock count and reset it)
The feature pack firmware (FP1) removed the low octane maps, and added flex fuel, launch control and TIM (total injection multiplier)

Inadvertently with the feature pack firmware, it shifted the memory locations used for knock reporting, so the knock reporting I had added longer functions (including for CA18DET) with the FP1 firmware. However I am working on a FP1.1 firmware which will do the following
- Anti stall code (limit TP when at a low throttle position to prevent overfuel and stall)
- Crank low voltage enrichment (some ECUs particularly NEO are hard to crank at low battery voltage with large injectors)
- Knock reporting by cylinder (replacing the older knock count) with per cylinder feedback
- Later model ECUs will have a longer consult stream with all the flex, duty cycle etc information packed in for faster comms
I mean, what if the detonation detection board reports to the main ECU board, that everything is ok, but the board self cannot detect knock anymore for some reason? Then no DTC would be set and no actions would be taken.
Incorrect. All knock reporting and detection is still as normal, and DTCs are still set
Only the reporting to Nistune (knock window) is not available with FP1 firmware
How could it display knock if the knock count is accidentally disabled in some boards?
It doesn't. I tested with 36F0F today and there is no knock count (displayed in the software) during testing. Same for 39F0F. When we added FP firmware I tested all the new functions but did not regression test this function so it is not available :oops:
Could you please explain how to enable or disable knock cells? What does that mean?
This is not applicable to CA18DET. However it is with later ECUs. It is covered in our Knock document here:
http://nistune.com/docs/KnockGuideSR20.pdf

This document will be updated once FP1.1 and FP2.1 are available with the new knock reporting (still writing the firmware currently)
In the context menu of the timing map I have "Highlight Knock Region C" disabled
This is only available where Region C maps have been located and identified. There are over 50 types of ECUs. I've only added them for a few basic ones during initial development (R32, BNR32, Z32, R34)
"Auto update knock maps", which confuses me again, since you say, knock maps have been removed
Yes they have. That should only really be enabled for non-FP firmware. I've fixed this now for the next release to hide when FP firmware loaded
My ECU number and the base image version are mentioned in my first post: 23710-39F00, Type 1 V5 board, 39F0F base image. Is that what you need?
Yes thanks so you have S13 CA18DET EDM 39F0F firmware
I've heared that the old ECU has some limitations like it cannot detect knock above 4000 rpm, just because of the lack of computing power. That's why I'm asking...
2020-05-17 18_59_58-Window.png
(48.03 KiB) Downloaded 2041 times
Good question. I've just tried it here and we can see that simulating knock on this ECU upto 3600rpm retards timing. After that there is no retard of timing. It may be due to engine noise? The knock retard limit RPM table goes upto 4000rpm which matches what you say
And because I am missing an answer and to be sure: The whole Flex Fuel area - can this be ignored by me if I do not use the Flex Fuel feature?
Correct. If you are not using flex, then you don't use the flex maps or tickbox to enable it
Another thing: Is it normal that the green selection stays where it is in the Knock Retard Limit table? Even though I'm moving around in the log? Shouldn't it be moving, too according to rpm and load?
Yes it should. The multigrid table doesn't appear to be using maptrace updates (only the initial trace when window opened). I'll look into that

Finally... you will see here the ECU retarding timing in my log as I put more and more knock into the ECU. Purple line is ignition timing
2020-05-17 19_04_16-Window.png
(25.36 KiB) Downloaded 2041 times

Re: Finally understanding knock / detonation detection on ca18det

Posted: Sun May 17, 2020 9:28 pm
by Matt
Another update. All maptraced tables have either X scale (single row table) or X,Y scale (map)

Knock retard tables don't have either, but have individual constants which the table references. That is why the scaling did not update. I've manually coded into use the indexes for TP and RPM comparing against constants for the next version
2020-05-17 19_55_48-Window.png
(9.11 KiB) Downloaded 2038 times

Re: Finally understanding knock / detonation detection on ca18det

Posted: Sun May 17, 2020 10:41 pm
by Matt
I've done the release today with those updates (since the vehicle list was missing on the previous release so had to hot fix that)

Re: Finally understanding knock / detonation detection on ca18det

Posted: Tue May 19, 2020 10:43 pm
by Shaker
Thank you, Matt, for your extraordinary support here and all the detailed explanations.

I'll work through this input soon.

One question: The firmware, you are working on - will this be flashed by the (future) nistune version during a normal "upload" to the ECU or would the board have to be reprogrammed?

Re: Finally understanding knock / detonation detection on ca18det

Posted: Fri May 22, 2020 11:44 am
by Matt
Unfortunately they all have to be reprogrammed. The board is too small sizewise to add another processor to do upload/download external flashing, and the ECU is running the program during upload / download of maps. It cannot change the program whilst it is running it... so the board needs to be removed and reflashed in an external programmer

I can do this for you no charge for the reflash of just the board (just cover return postage), but after the COVID has stopped wreaking havoc on our mail systems... since its taking upto 2 months for packages to get in/out of Australia

Re: Finally understanding knock / detonation detection on ca18det

Posted: Tue May 26, 2020 6:51 pm
by Shaker
That's very kind of you. I'm afraid, since the S13 is my daily I cannot remove the board for so long. Maybe there is another solution? Local dealers that can flash the board? Will they be supported with the new firmware?

But if the S13 doesn't detect knock above 4000 rpm, I might want to look for a different solution here. Thus I may not need the update of my board. We'll see. Depends on which other changes will have been made that I'd like to profit from.

Re: Finally understanding knock / detonation detection on ca18det

Posted: Wed May 27, 2020 9:50 pm
by Matt
Yes local dealers can reflash the board once the new firmware is available

Re: Finally understanding knock / detonation detection on ca18det

Posted: Fri Mar 26, 2021 1:42 am
by Nenaudelis
Matt wrote: Sun May 17, 2020 8:35 pm
I've heared that the old ECU has some limitations like it cannot detect knock above 4000 rpm, just because of the lack of computing power. That's why I'm asking...
2020-05-17 18_59_58-Window.png
Good question. I've just tried it here and we can see that simulating knock on this ECU upto 3600rpm retards timing. After that there is no retard of timing. It may be due to engine noise? The knock retard limit RPM table goes upto 4000rpm which matches what you say
Is this also applies to other ECU's, for example SR20?
It meant that it is nonsense to have knock enabled, because it wouldn't help anyway. Because now mostly of engines are heavy upgraded with big turbos, which ones starting to spool over 4000rpm and most likely knocking later on.

Re: Finally understanding knock / detonation detection on ca18det

Posted: Wed Mar 31, 2021 4:30 am
by Matt
It's more of a case of noisy engine internals (rattles etc) than boost from the larger turbo that may affect false positives and knock retard

SR20 maps are the same, and better defined than CA18DET in our address definitions

Re: Finally understanding knock / detonation detection on ca18det

Posted: Tue Apr 20, 2021 12:01 am
by Run_Stop_Restore
Hi Matt,

sorry to Hijack this thread but how far are you with the Firmware update that will enable Cyl Selective Knock Logging?

Thanks in advance
Felix